4 Slope Fields, Solution Curves, and Euler’s Method You may recall from calculus that it is not always possible to evaluate an integral symbolically. For example, the integral Z Here the term elementary functions refers to polynomials, exponentials, logarithms, trigonometric functions, inverse trigonometric functions, and combinations thereof. Of course, y 0 = sin x2 is itself a differential equation whose solutions can’t be written using elementary formulas. sin x2 dx cannot be expressed as a formula involving elementary functions—there simply isn’t a formula whose derivative is sin x2 . Of course, this doesn’t mean that we can’t integrate sin x2 . The antiderivative still exists. We can graph it, and we can evaluate it numerically using approximate methods. We just can’t write a formula for it that doesn’t involve an integral. The same thing happens with differential equations: even if a solution exists, there might no be way to express it as a formula. For example, the equation y 0 = x3 + y 3 has plenty of solutions, but none of these solutions can be written as a formula involving elementary functions. If we want to solve this differential equation, we must resort to either numerical or graphical methods. Explicit Equations But how can we possibly know that a differential equation has a solution if we can’t find a formula for it? Well, consider again the equation y 0 = x3 + y 3 . There is a simple intuitive argument that this differential equation must have a solution. Imagine an object moving vertically, whose vertical position y changes with time. In this case, the equation This is the same as the last equation, except that we have renamed x to t. dy = t3 + y 3 dt can be thought of as instructions for how to move the object. Specifically, this equation says what the velocity of the object should be in terms of the time t and the vertical position y. If we just move the object according to these instructions, we will obtain a position function y(t), which must be a solution to this differential equation. This same reasoning applies to any differential equation of the form y 0 = F (x, y), Almost every differential equation that arises in science is explicit. where F (x, y) is any expression involving x and y. A differential equation of this form is said to be explicit, and can be thought of as a set of instructions for how quickly y should change in terms of x and y. Intuitively, we expect that any explicit equation should have at least one solution. Indeed, we expect that any initial value problem of the form y 0 = F (x, y), y(a) = b should have a uniquely determined solution. The following theorem makes this intuition rigorous. SLOPE FIELDS, SOLUTION CURVES, AND EULER’S METHOD 2 EXISTENCE AND UNIQUENESS OF SOLUTIONS Consider an initial value problem of the form y 0 = F (x, y), y(a) = b. Assuming the partial derivatives of the function F exist and are continuous, this initial value problem has a uniquely determined solution. In case you aren’t familiar with partial derivatives (which are normally covered in a vector calculus course), the important thing to understand here is that the initial value problem y 0 = F (x, y), y(a) = b. usually has a uniquely determined solution, and there’s a fairly simple test we can use to make sure. Slope Fields and Solution Curves 1.5 One way to understand a differential equation is to draw graphs of its solutions. These graphs are curves in the xy-plane, and are known as solution curves. There is a nice way to figure out the geometry of the solution curves directly from the differential equation. For example, consider the equation: 1.0 y 0 = (x − 1)2 − y 2 . 0.5 0.5 1.0 1.5 2.0 2.5 Figure 1: Slope field for the equation y 0 = (x − 1)2 − y 2 . We can interpret this equation as a formula for the slope y 0 of the solution curve in terms of x and y. For example, if a solution curve for this equation goes through the point (4, 2), its slope at that point must be 5, since plugging in x = 4 and y = 2 into the differential equation gives y 0 = 5. We can visualize this information by drawing small line segments with the appropriate slopes at various points in the xy-plane, as shown in Figure 1. This picture is called a slope field, and was drawn using the following procedure: 1.5 1. First, an array of x and y values were chosen. In this case, the x-values were 0.0, 0.1, 0.2, . . . , 2.5, and the y-values were 0.0, 0.1, 0.2, . . . , 1.5. 1.0 2. For each (x, y) pair, the slope y 0 was computed using the equation y 0 = (x − 1)2 − y 2 . 0.5 0.5 1.0 1.5 2.0 2.5 Figure 2: Three solutions to this equation. 1.5 1.0 0.5 0.5 1.0 1.5 2.0 2.5 Figure 3: The solution for which y(0) = 1. 3. At each point (x, y), a small line segment was drawn with the computed slope. We can think of a slope field as a picture of a differential equation, in the sense that it contains the same information that the differential equation does. The solutions to a differential equation are curves that follow the slopes of the slope field, as shown in Figure 2. By the way, we can understand an initial condition graphically as a point on the plane that the solution curve must go through. For example, the solution to the initial value problem y 0 = (x − 1)2 − y 2 , y(0) = 1 is shown in Figure 3. As you can see, the initial condition y(0) = 1 corresponds to the fact that the solution curve goes through the point (0, 1). Intuitively, the solution to this initial value problem can be obtained by starting at the point (0, 1) and then “following the slopes” to obtain a solution curve. SLOPE FIELDS, SOLUTION CURVES, AND EULER’S METHOD 3 EXAMPLE 1 Recall that the logistic equation is the differential equation dP P = kP 1 − dt Pmax P Pmax t Figure 4: Slope field and solution curves for the logistic equation. where k and Pmax are constants. This equation can be used to modeled the growth of a population in an environment with a finite carrying capacity Pmax . A typical slope field for the logistic equation is shown in Figure 4. Several solutions to the equation are also shown. Note that the slopes point up if P < Pmax and down if P > Pmax , so in either case the population approaches the carrying capacity. (The case where P > Pmax corresponds to a situation where the population is too high to be sustainable.) By the way, observe that the slopes for this equation are the same on each horizontal line. This is because the formula for dP/dt depends on P only, with no explicit dependence on the variable t. Based on the pictures so far, you might have noticed a few important properties of the solution curves for a differential equation. For example, Figure 5 shows several different solution curves for the equation 1.4 1.2 1.0 y 0 = (x − 1)2 − y 2 . 0.8 0.6 0.4 These curves have the following properties: 0.2 0.0 0.0 0.5 1.0 1.5 2.0 2.5 Figure 5: Solution curves for the equation y 0 = (x − 1)2 − y 2 . 1. The curves entirely fill the plane, in the sense that each point of the plane has a solution curve going through it. This is because any point in the plane can be thought of as an initial condition, which means that it lies on the solution curve for the corresponding initial value problem. 2. No two of the curves intersect, i.e. each point on the plane is contained in only one curve. This is because of the uniqueness of solutions—each point can be thought of as an initial condition, and there should only be one solution corresponding to a given initial condition. These properties are typical for the solutions to explicit first-order differential equations. Euler’s Method Euler’s method is a simple numerical method that can be used to approximate the solutions to explicit first-order equations. Given an initial value problem y 0 = (x − 1)2 − y 2 , y(0) = 0.5, the idea is to estimate the values of y and y 0 at several successive values of x: x 0.0 0.5 1.0 1.5 2.0 2.5 y 0.5 ? ? ? ? ? ? ? ? ? ? ? y The choice of ∆x = 0.5 is arbitrary. As we shall see, choosing a smaller value of ∆x would improve accuracy but also make the computation longer. 0 As you can see, we have chosen to use values of x that are spaced ∆x = 0.5 apart. First, we compute the initial value of y 0 by plugging x = 0.0 and y = 0.5 into the differential equation: y 0 = (0.0 − 1)2 − (0.5)2 = 0.75. SLOPE FIELDS, SOLUTION CURVES, AND EULER’S METHOD 4 We fill this value into the table: x 0.0 0.5 1.0 1.5 2.0 2.5 y 0.5 ? ? ? ? ? 0.75 ? ? ? ? ? y 0 Next we use a linear approximation to estimate the next value of y. That is, we use the computed value of y 0 to estimate the slope ∆y/∆x: ∆y ≈ y0 ∆x Be careful not to confuse ∆y with y. After we figure out that ∆y ≈ 0.375, we must add this to the old value of y (which was 0.5) to get the new value of y (namely 0.875). ∆y ≈ 0.75. 0.5 so Solving gives ∆y ≈ 0.375, and we add this to the old value of y to get the new value: x 0.0 0.5 1.0 1.5 2.0 2.5 y 0.5 0.875 ? ? ? ? 0.75 ? ? ? ? ? y 0 Now, because x and y have changed, the value of y 0 has also changed. We compute the new value of y 0 y 0 = (0.5 − 1)2 − (0.875)2 ≈ −0.5156 and fill it in: x 0.0 0.5 1.0 1.5 2.0 2.5 y 0.5 0.875 ? ? ? ? 0.75 −0.5156 ? ? ? ? y Be careful not to confuse ∆x with x. Although each column has a different x value, the value of ∆x is 0.5 for this entire computation. 0 The process just repeats from here. We have ∆y ≈ y0 ∆x ∆y ≈ −0.5156. 0.5 so which gives ∆y ≈ −0.2578. Adding this to the previous value of y (the 0.875) gives 0.6172, which we fill into the table: x 0.0 0.5 1.0 1.5 2.0 2.5 y 0.5 0.875 0.6172 ? ? ? 0.75 −0.5156 ? ? ? ? y 0 We repeat this process three more times to finish off the table: Since the goal is to estimate values for y, there is no need to compute the value of y 0 in the last column. x 0.0 0.5 y 0.5 0.875 0.75 −0.5156 y 0 1.0 1.5 2.0 2.5 0.6172 0.4267 0.4607 0.8546 −0.3809 0.0679 0.7878 — EULER’S METHOD Euler’s method involves repeating the following process, starting with the initial values of x and y: 1. Use the differential equation to compute the value of y 0 . ∆y 2. Use the approximation y 0 ≈ to estimate ∆y. ∆x 3. Add ∆y to the old value of y to obtain the new value of y, and then go back to step 1. SLOPE FIELDS, SOLUTION CURVES, AND EULER’S METHOD 5 Further Numerical Methods Euler’s method is only the simplest method for approximating solutions to differential equations. Finding better ways of approximating solutions is an ongoing subject of research, and is one of the primary questions in the field of mathematics known as numerical analysis. One basic improvement is to estimate the slope of each straight segment using a combination of y 0 values corresponding to different values of x. This idea leads to a set of possible methods known collectively as Runge-Kutta methods. Another possible improvement, often combined with Runge-Kutta, is to change the step size ∆x adaptively, using small steps when the value of y 0 seems to be changing quickly, and using large steps when the slope seems roughly constant. Modern computer algebra systems such as Mathematica and Sage have a variety of different numerical methods built into the system. Though you can choose a method explicitly to solve a given problem, most computer algebra systems also have a built-in algorithm to choose an appropriate method based on the properties of the given equation. 1.5 1.0 0.5 0.5 1.0 1.5 2.0 2.5 Figure 6: Euler’s method with ∆x = 0.5. By the way, this method should remind you of numerical methods for computing integrals by adding up the areas of vertical rectangles. Indeed, in the special case where the differential equation has the form y 0 = f (x) 1.5 this method is essentially identical to the Left Endpoint Rule for approximating an integral. In general, though, Euler’s method works a bit differently than a Riemann sum because of its recursive nature: we must use the results of each column to compute the numbers in the next column. 1.0 0.5 The Geometry of Euler’s Method 0.5 1.0 1.5 2.0 2.5 Figure 7: Euler’s method with ∆x = 0.25. When we use Euler’s method, we are essentially approximating the solution curve by a straight line between successive values of x. For example, Figure 6 shows Euler’s method for the example above, i.e. the initial value problem y 0 = (x − 1)2 − y 2 , 1.5 1.0 0.5 0.5 1.0 1.5 2.0 2.5 Figure 8: Euler’s method with ∆x = 0.01. y(0) = 0.5 with ∆x = 0.5. The six marked points are the points that we computed, and the straight lines represent the linear approximations that we used for the estimation. Note that the slope of each straight line is correct on its left end, but becomes less and less correct as we move right towards the next x-value. We can improve the accuracy of Euler’s method by using a smaller step size. For example, Figure 7 shows Euler’s method applied to the same initial value problem using a step size of ∆x = 0.25, and Figure 8 shows a step size of ∆x = 0.01. For the step size of 0.01, all of the y-values are accurate to within 0.004, which is about a fourth of the width of the red line that we’re using for the graph. EXERCISES 1. Consider the following initial value problem: y 0 = x2 − y 3 , y(0) = 1. Use Euler’s method with a step size of 0.2 to estimate y(1). 2. Consider the following initial value problem: y 0 = xy − 2, y(2) = 1. Use Euler’s method with a step size of 0.5 to estimate y(4). SLOPE FIELDS, SOLUTION CURVES, AND EULER’S METHOD 3. A ball is dropped from the top of a tall building. If air resistance is taken into account, the downward velocity v of the ball is modeled by the differential equation dv = g − kv 2 dt where g = 9.8 m/s2 , and k = 0.1/m. Assuming the initial velocity of the ball is zero, use Euler’s method with a step size of 0.25 seconds to estimate the velocity of the ball after one second. 4. The following picture shows the slope field for a certain explicit first-order equation 2.0 6 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0.0 0.0 0.0 0.5 1.0 1.5 2.0 0.0 0.5 (III) 1.0 1.5 2.0 (IV) 6. y 0 = xy 7. y 0 = x2 + y 2 8. y 0 = x2 − y 2 9. y 0 = y − x2 10–13 Match each differential equation with the corresponding solution curve. 1.5 3.0 3.0 2.5 2.5 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 1.0 0.5 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.0 (a) Use this slope field to sketch the solution curve satisfying y(0) = 1. 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.0 0.5 1.0 (I) (b) Given that y(0) = 1, use your answer to part (a) to estimate the value of y(3). 5. Consider the following differential equation. −1 if x < 1, y0 = 0 if 1 ≤ x ≤ 2, 1 if x > 2. (a) Sketch the slope field for this equation. Your sketch should include the range 0 ≤ x ≤ 4 and 0 ≤ y ≤ 3. 3.0 3.0 2.5 2.5 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.0 0.5 1.0 (III) 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0.0 0.0 0.0 0.5 1.0 (I) 1.5 2.0 0.0 0.5 1.0 (II) 1.5 2.0 1.5 (IV) (b) Given that y(0) = 1.5, find the value of y(4). 6–9 Match each differential equation with the corresponding slope field. 1.5 2.0 2.5 3.0 2.0 2.5 3.0 (II) 10. y 0 = sin(xy) 11. y 0 = x − y 3 12. y 0 = e−xy 13. y 0 = cos(xy) Answers x 1. y 0.0 0.2 1 0.8 0.4 0.6 0.8 1.0 −1 −0.472 −0.1913 0.0628 0.3257 x 2.0 2.5 2. y 1 y0 0 1 3.0 3.5 4.0 1.25 2.125 4.8438 0.5 1.75 5.4375 t 0.00 0.25 v 0 2.45 dv/dt 9.8 3. 4. (a) so y(1) ≈ 0.745. 0.7056 0.6673 0.6799 0.7450 y0 — so y(4) ≈ 4.844. — 0.50 0.75 1.00 so v(1) ≈ 7.985 m/s. 4.7499 6.6359 7.9850 9.1998 7.5438 5.3965 2.0 — (b) y(3) is somewhere between 0 and 0.5. 1.5 1.0 0.5 0.5 1.0 1.5 2.0 2.5 3.0 3.0 5. (a) 3.0 (b) 2.5 2.0 2.0 1.5 1.5 1.0 y(4) = 2.5 2.5 1.0 0.5 0.5 0.0 0 6. III 7. IV 1 2 8. II 3 9. I 4 10. III 1 11. II 12. IV 2 3 13. I 4